You know, for most of my life I thought my tendency to go ‘all in’ on whatever (or whoever) caught my interest was just a random personality trait, but like many other things about me, it turns out to be a feature of autism! For those of you who don’t experience it yourselves, this is what it looks like when an Autistic Fascination takes over your entire life.
The day after I put down my first two buttons, when it became clear that Tashi was a quick study and I was going to have at least one talking pet in the house, I started the serious researching and planning. I read every web site and news article I could find, plus almost all of the bulletin board and several months of history in the Discord group. Normally I never look at video social media, but now I hunted down every account I could find featuring pets with extensive soundboards, and watched thousands of videos, making a list of the words they used, as well as keeping track of what words Jak and I tended to use with our pets already and trying to anticipate which words they might find useful.
I arranged the list into six groups (as there are six different hextile colors), and kept reorganizing and tweaking the plan as I learned new things. Once I was pretty sure of how I’d handle grouping, I went into my vector drawing software and drew out scale versions of hextiles with button holes in the correct positions, and started testing layouts.
By two and a half weeks AB (after buttons), when I decided I would eat the extra cost of shipping to Mexico and placed the order for expansion equipment, I used my soundboard plan to inform my choice of tile numbers and colors. But it’s a dynamic document; it stays open on my desktop and I make changes about once a day on average, as I have new inputs, insights, or ideas.
It’s changed a lot since the third week of September. Here’s the earliest snapshot I saved out, from October 10 (five and a half weeks AB):
Everything is to scale: the top and left edges are walls, and the right and bottom edges basically delimit the maximum space that we can possibly devote to a soundboard in our living room. The black square in the upper left corner is the subwoofer (which you can see below the wall-mounted TV in the photo in my last post). The bright tiles are the three I’d started with plus the nine more on the way; the opaque pastels are additional ones I was at that point expecting to need eventually. White buttons are the nine words I had already introduced on the starter board; the semi-opaque buttons mark words that I thought (50-100% certainty) Tashi already knew from our verbal modelling; while the faintest circles are words I didn’t think very likely Tashi understood yet. In the white space were other words I was considering but wasn’t sure of yet — the more I learned about other button pets, the larger my concept of what was possible became. You can see here that I had only recently cottoned on to the usefulness of body parts (they can tell you where it hurts) and adjectival pairs.
I had just realized the day before this snapshot that I was going to need more verbs, more quickly than I’d originally expected, and so I had rearranged the color sequence so that actions would be on yellow tiles (that I would soon have three of) rather than purple (that I would have only two of). The color sequence — which tiles were next to each other — was tricky; it needed to be pleasing to human eyes, but also distinctive to dog vision. Dogs see fewer colors than humans do; so I converted the current FluentPet tile colors and checked various arrangements. Orange and yellow were likely too similar to dog eyes, so I made sure to keep them far apart.
(Strangely enough, we still don’t know whether cats are dichromats like dogs or trichromats like humans. But I figured at worst they see something similar to dogs, and at best something closer to what we do, so I figured if I made it work for both dog and human vision — and also made sure the cyan tiles were separate from the red ones — the cats would be okay.)
Here’s my final color test, showing dog view of the twelve tiles I would have when the expansion arrived:
Nothing is up against the wall, because one of the things I’d learned from the button-teacher community is how much ease of access matters — hard to reach buttons often don’t get used. I iterated and iterated to make sure that every button was no more than two deep from a potential standing place. However, as time went on and my list of important words grew, I decided to shrink the wall alleyways a little bit in order to fit a couple more tiles in.
After a lot of delay and frustration that I won’t recount here, I finally got my upgrade equipment from FluentPet and was able to start using it on October 21. Here’s the plan as it stood then:
You can see that not only had I added a fifth ‘actions’ tile and a second ‘places’ tile into the long-term plan, but a whole island in the center area, for those body parts and non-feeling adjectives. I left those grey, because I don’t yet know what colors I’ll use there. In late September FluentPet announced that they are going to add some tiles in shades of grey and brown (presumably in response to the people I’ve seen complain about the way rainbow jewel tones clash with their decor); if they’re ready in time I’ll probably use two of those colors. Otherwise, I’ll have to reuse two of the colors I’ve already got, or figure out a way to paint the tiles.
I actually held back the second blue tile for use in target-training Rikki, and just put down the one for our family names, so what I had to start was three separate sections: five, three, and three. I soon found out that three-tile sections are lightweight enough that they get skewed around all over the place under the attentions of a rambunctious
16 17 18-kilo puppy, and it wasn’t too long before I attached the red and purple tiles at their closest points to keep that from happening. So then I had two sections: five and six.
As far as words go … I had a hard time both figuring out how to prioritize and holding myself back. We’d been waiting a month to add more buttons, and Tashi had picked up a lot of words.
I wanted to see if I could fix the off-label SKRITCHES problem, so I gave him the cluster of five family names right away, as well as WANT and HELP a few hours later. We’d been verbally modelling “want” a lot, “help” only a little, so I wasn’t sure what to expect. Tashi wasn’t the least bit dismayed by all this change; in fact, he started using HELP and WANT right away just as though he’d always had them. And sure enough, the off-label use of SKRITCHES stopped immediately, which I take as implicit confirmation that he just didn’t have access to enough words before.
It wasn’t clear to me how much he grokked questions — sometimes he would respond, either with buttons or body language; other times he would just stare at me, or ignore me completely. Knowing that buttons can sometimes put extra focus on a word (a word that you can say is more interesting than a word that is only ever said to you), I decided to add WHAT? on day two of the new layout. The way he used it at first left me suspecting that he might be confusing WHAT? with WANT, despite the fact I’d added a slight tonal uplift to the former. But after a couple-three days that sorted itself out, and he stopped using the WHAT? button in places where WANT made more sense. I think maybe having two different buttons helped him understand that they were two different words verbally, too? But as with many things, that’s just an informed guess.
I also added TUG at the same time — a Very Important Word from Tashi’s perspective, and one that I was 100% certain he knew. So in less than two days, we’d gone from nine buttons to eighteen, and Tashi was enthusiastically there for it.
I gave it three days, and then pushed forward again. Still working on questions, I gave him HMM?, which is like a verbalized question mark — the common way among button teachers of marking yes-no questions. So far — excepting the first couple of days when I think he was confusing WANT/WHAT? — he very rarely uses either question button himself. (Of course, his options are still pretty limited. Eventually I expect to give him WHERE?, WHEN?, WHY?, and WHO?) I also put down MORE, which is a word I verbally use with him a lot and I thought for sure he would want to say often … but to my surprise it’s relatively rare. When he does, though, it’s always in perfect context, so that’s one I’m sure he knows exactly what it means.
After that, I started in on feelings. Being very aware of the need to conserve on buttons, I had gone back and forth for weeks about whether to separate or conflate the ideas of ‘frustrated’ and ‘mad’; eventually I ended on separate. Frustration is a Very Big Emotion for Tashi, and I’d been modelling it the longest, so FRUSTRATED was his first emotion button.
FRUSTRATED, not too surprisingly, instantly became one of Tashi’s most-used words. After a few more days I added MAD and HURT, and then a couple of days later, BORED and EXCITED. EXCITED was easy to model; I deliberately added it (and BORED) on the morning before his weekly playdate with my friend’s dog. All I had to do was tell him that “Chewie is coming soon to play outside”, and suddenly he was bounding around like a maniac. “Tashi is excited!” I tell him. TASHI EXCITED. Boom, done.
I’d already been verbally modelling “bored” every time he would go to the buttons and do a random assortment of Press All The Things; I used “mad” mostly to refer to myself when he would bite me or harass the cats, Gracie when she was hissing and spitting and swatting him after said harassment (Rikki just runs away, causing Tashi to chase), and himself when his lip snarl told me his frustration was at its worst. “Hurt” I’ve only been able to model with physical wounds — a cut on my finger, a raw spot on my heel — but I also used it (sometimes with “mad”) when he would bite me hard, or I heard one of the cats yelp in a way that suggested pain. I’ve seen multiple people say that their button learners generalized from physical pain to emotional pain on their own, but I haven’t had the chance to model it that way yet.
Four of the five emotion words immediately got super-frequent use — the only one he rarely touched was MAD (and he was still using it, just more like a couple times over a week rather than multiple times a day). I was not expecting this at all — if you’d told me three months ago that my dog would be regularly telling me about his emotions, I would have given you a strong side-eye. For about a week I was doubting that he actually had any idea of the intended meaning of HURT, because he paired it with such odd things and under odd circumstances, but several things have clicked for me over the last few days and now I think he probably does. (More on that in next post.) Interestingly, with the introduction of emotion words, WANT and HELP are getting less use.
Based on body language, I noticed Tashi actually uses EXCITED without the English connotation of “happy anticipation”, more like what I would call “energetic” or “riled up”. Which I’ve decided is both reasonable and useful; I’ve already started adopting his meaning when talking to him. A HAPPY button was coming, so we could use the HAPPY / EXCITED combo for “happy anticipation.”
In the middle of all these new buttons, I made some changes to the long-term plan that made it necessary to tweak the button layout of the blue and purple tiles. Plus, attaching red and purple together to keep those sections from scooting around all over the place meant that when I put in MAD it went right next to FOOD, which was suboptimal from my extremely meticulous perspective. So I ended up slightly adjusting the whole right-hand six-tile section, and held my breath to see whether it threw him for a loop or not.
Luckily for me, the answer was ‘not’. He visibly startled a couple of times when he pressed a button and it made a different sound than he expected, but within a few hours he had sorted it out and was seemingly unfazed.
Non-emotion words: along with HURT and MAD I had also given him SOON and CHASE. SOON … I’m not sure he’s ever actually used SOON on his own, though I started verbally modelling it weeks before the expansion arrived. I honestly still have no clear sense whether he understands what it means when I say it or not. CHASE was a 100% for-sure word, Tashi’s other favorite type of PLAY along with TUG, plus it’s doubly useful because it’s one of the words I use all the time in the context of Leave The Fucking Cats Alone, Dog: TASHI NO CHASE RIKKI. (BOTHER and BITE are the other ones, more often applied to Gracie, who will complain, warn, and fight in that order, but rarely run. He’s getting better; he’ll break away about half the time I call him off a cat now, instead of … never.)
The same day as BORED and EXCITED — Halloween, in fact — I gave him SETTLE, which was another 100% known word, because it had been so important in training him into some kind of (partial) self-control (another ongoing project). I also added YES and NO, because 1) it seemed like he ought to have a way to negate things, and 2) I kept realizing how useful it would be if he could learn to answer yes/no questions with an actual YES or NO. I would imagine he has some ideas about what those words mean, but on the other hand I am aware that with pets we use “yes” and “no” as approval and admonishment, which can make it hard for them to start using them to answer questions.
In fact — Tashi is my ninth pet, and I have never had a dog or cat who has ignored verbal “no” like he does. I mean, as far as I can tell, Rikki only understands two words, and they are “Rikki” and “no”! I cannot imagine, given the way Tashi has picked up other words, that he doesn’t understand what (for example) “no bite” means, but you wouldn’t know that from his behavior. It seems more likely he just gives zero fucks.
We have tried everything short of physical aggression to curb the biting. Jak, who is much bigger, louder, and deeper-voiced than I am, got a little respect from yelling at him, but with me it just made him double down. I yelped like puppies do, to let him know he’d hurt me, but he didn’t care. Muzzle grabs are supposed to be what adult dogs do to tell puppies they’ve overstepped, but they just made him more mad. Squirting him in the face with water caused him to back off for a few days, and then it stopped surprising him and just started pissing him off, so I had to abandon that. For a long time the only thing that even sometimes worked was to firmly command him to “down”. It would usually take a few repetitions, and didn’t always work, but it was the best thing I had.
To my surprise, though, since giving him the MAD button, he’s begun showing respect for “Karawynn mad.” I don’t yell it, just say it in a harsh tone of voice (and also only when I actually mean it at least a little). And more and more, that causes him to back off … at least from me. Not as often from the cats. I say “Gracie mad” a lot too, and if that would get the same respect, we’d be in better shape. (She’s a fierce little thing, and protective; he’s now four times her size, but when he’s biting and I make hurt-noise responses, either involuntarily or deliberately, she will often come running in and get between Tashi and me, or even go into whirlwind attack.)
Anyway, I can’t ever be entirely certain how well he understands the more abstract words; even with the emotions, I have only circumstantial evidence — it’s good enough for me, but it’s not hard scientific proof. I don’t think I’ve been pushing him too hard — in some ways I’m holding him back, because I can’t get all the words he knows on the board fast enough. At least when it comes to Button Pidgin, it’s looking more and more like Tashi is near the far end of the bell curve. Every word added opens up multiple possibilities, and in the last week we’ve had what I would consider our first real back-and-forth conversation.
But back to a week ago: we were up to thirty-one buttons on the board, and I believed that Tashi already understood, at a minimum, twenty-five of them well. Here’s a photo (click through to see an enlarged version if you want to read the labels):
The Connect speaker sits nicely on top of the subwoofer … at least until I reach a hundred buttons and it’s maxed out and I have to buy a second one. Which suddenly doesn’t seem that far away, given that I’d bought forty-eight buttons expecting that to last us three months, and eleven days later I was down to seventeen. At that pace I would run out of buttons on November ninth.
The one button on the board I didn’t list above is the one that wasn’t intended at all for Tashi. While all the rest of this was going on, I’d also been doing target training sessions with Rikki every day or two — perhaps five or six of them by October 30. Which is when (after rearranging the right-hand tiles) I decided to put down CATNIP so I could model it for Rikki, in hopeful anticipation of his eventual transition to the real board.
That had a number of unforeseen but ultimately fascinating repercussions, which I will describe in the next installment.
I need more buttons, more hextiles, and a couple of dedicated cameras so I can start collecting video evidence — all of which is a lot more expensive to get in Mexico than in the US. If you happen to find Tashi’s progress half as fascinating as I do, and you’d like to help me keep up with him, you can contribute to the project with either a FluentPet gift card or a Ko-fi donation.